Google Open-Sources PointDiT: Minimalist ViT for Monocular 3D Reconstruction
CSProfKGD · x · 2026-08-25
Google Research has released PointDiT (ICML 2026), a pixel-space Diffusion Transformer for monocular geometry estimation. The project is fully open-sourced, including code, pre-trained models, and the full training/evaluation pipeline.
PointDiT features a minimalist design: operating directly on raw patches to estimate dense 3D point maps using a plain ViT. It eliminates the need for hybrid ViT+Conv architectures, lossy VAEs, or complex training losses, relying instead on conditioning on frozen pre-trained DINOv3 image tokens.
More from Research
- Pathway CSO pitches BDH as post-Transformer architecture uniting memory, reasoning, adaptation — schwentker · 2026-10-15
- EmbodiedEvo tops RoboDojo at 38.95% success by letting robots inherit failed practice — thetripathi58 · 2026-10-15
- 797GB of n-gram counts in CPU RAM let 1B models match 57% more compute — PangWeiKoh · 2026-10-15
- Google's HARNESS² evolves agent harnesses without verifiers, up to +11.5 points on WorkBuddy-Bench — dair_ai · 2026-10-15
- Subbarao Kambhampati to dissect the 'mythos of LLM thinking tokens' at UMN — rao2z · 2026-10-15
- How 3D Gaussian Splatting gives robots spatial memory of what's out of view — TheTuringPost · 2026-10-15